src/Pure/General/symbol.scala
author wenzelm
Mon, 11 Jan 2010 21:37:48 +0100
changeset 34319 f879b649ac4c
parent 34193 d3358b909c40
child 36035 3ff725ac13a4
permissions -rw-r--r--
clarified Symbol.is_plain/is_wellformed -- is_closed was rejecting plain backslashes;
wenzelm@27901
     1
/*  Title:      Pure/General/symbol.scala
wenzelm@27901
     2
    Author:     Makarius
wenzelm@27901
     3
wenzelm@27924
     4
Detecting and recoding Isabelle symbols.
wenzelm@27901
     5
*/
wenzelm@27901
     6
wenzelm@27901
     7
package isabelle
wenzelm@27901
     8
wenzelm@27918
     9
import scala.io.Source
wenzelm@31939
    10
import scala.collection.{jcl, mutable}
wenzelm@31537
    11
import scala.util.matching.Regex
wenzelm@27901
    12
wenzelm@27901
    13
wenzelm@31537
    14
object Symbol
wenzelm@31537
    15
{
wenzelm@34007
    16
  /* Symbol regexps */
wenzelm@27901
    17
wenzelm@31537
    18
  private val plain = new Regex("""(?xs)
wenzelm@31537
    19
    [^\\ \ud800-\udfff] | [\ud800-\udbff][\udc00-\udfff] """)
wenzelm@27901
    20
wenzelm@31537
    21
  private val symbol = new Regex("""(?xs)
wenzelm@31548
    22
      \\ < (?:
wenzelm@27924
    23
      \^? [A-Za-z][A-Za-z0-9_']* |
wenzelm@27924
    24
      \^raw: [\x20-\x7e\u0100-\uffff && [^.>]]* ) >""")
wenzelm@27923
    25
wenzelm@34319
    26
  // FIXME cover bad surrogates!?
wenzelm@34319
    27
  // FIXME check wrt. ML version
wenzelm@31537
    28
  private val bad_symbol = new Regex("(?xs) (?!" + symbol + ")" +
wenzelm@31548
    29
    """ \\ < (?: (?! \s | [\"`\\] | \(\* | \*\) | \{\* | \*\} ) . )*""")
wenzelm@27923
    30
wenzelm@27939
    31
  // total pattern
wenzelm@31537
    32
  val regex = new Regex(plain + "|" + symbol + "|" + bad_symbol + "| .")
wenzelm@27918
    33
wenzelm@34146
    34
wenzelm@34146
    35
  /* basic matching */
wenzelm@34146
    36
wenzelm@34319
    37
  def is_plain(c: Char): Boolean = !(c == '\\' || '\ud800' <= c && c <= '\udfff')
wenzelm@34146
    38
wenzelm@34319
    39
  def is_wellformed(s: CharSequence): Boolean =
wenzelm@34319
    40
    s.length == 1 && is_plain(s.charAt(0)) || !bad_symbol.pattern.matcher(s).matches
wenzelm@34146
    41
wenzelm@34146
    42
  class Matcher(text: CharSequence)
wenzelm@34146
    43
  {
wenzelm@34146
    44
    private val matcher = regex.pattern.matcher(text)
wenzelm@34146
    45
    def apply(start: Int, end: Int): Int =
wenzelm@34146
    46
    {
wenzelm@34146
    47
      require(0 <= start && start < end && end <= text.length)
wenzelm@34319
    48
      if (is_plain(text.charAt(start))) 1
wenzelm@34147
    49
      else {
wenzelm@34146
    50
        matcher.region(start, end).lookingAt
wenzelm@34146
    51
        matcher.group.length
wenzelm@34146
    52
      }
wenzelm@34146
    53
    }
wenzelm@31537
    54
  }
wenzelm@27901
    55
wenzelm@27901
    56
wenzelm@34007
    57
  /* elements */
wenzelm@31939
    58
wenzelm@34146
    59
  def elements(text: CharSequence) = new Iterator[CharSequence]
wenzelm@34143
    60
  {
wenzelm@34146
    61
    private val matcher = new Matcher(text)
wenzelm@34007
    62
    private var i = 0
wenzelm@34007
    63
    def hasNext = i < text.length
wenzelm@34007
    64
    def next = {
wenzelm@34146
    65
      val n = matcher(i, text.length)
wenzelm@34146
    66
      val s = text.subSequence(i, i + n)
wenzelm@34146
    67
      i += n
wenzelm@34146
    68
      s
wenzelm@34007
    69
    }
wenzelm@34007
    70
  }
wenzelm@34007
    71
wenzelm@34007
    72
wenzelm@34007
    73
  /* decoding offsets */
wenzelm@34007
    74
wenzelm@34007
    75
  class Index(text: CharSequence)
wenzelm@31939
    76
  {
wenzelm@31939
    77
    case class Entry(chr: Int, sym: Int)
wenzelm@31939
    78
    val index: Array[Entry] =
wenzelm@31939
    79
    {
wenzelm@34146
    80
      val matcher = new Matcher(text)
wenzelm@31939
    81
      val buf = new mutable.ArrayBuffer[Entry]
wenzelm@31939
    82
      var chr = 0
wenzelm@31939
    83
      var sym = 0
wenzelm@34007
    84
      while (chr < text.length) {
wenzelm@34146
    85
        val n = matcher(chr, text.length)
wenzelm@34146
    86
        chr += n
wenzelm@31939
    87
        sym += 1
wenzelm@34146
    88
        if (n > 1) buf += Entry(chr, sym)
wenzelm@31939
    89
      }
wenzelm@31939
    90
      buf.toArray
wenzelm@31939
    91
    }
wenzelm@31939
    92
    def decode(sym: Int): Int =
wenzelm@31939
    93
    {
wenzelm@31939
    94
      val end = index.length
wenzelm@31939
    95
      def bisect(a: Int, b: Int): Int =
wenzelm@31939
    96
      {
wenzelm@31939
    97
        if (a < b) {
wenzelm@31939
    98
          val c = (a + b) / 2
wenzelm@31939
    99
          if (sym < index(c).sym) bisect(a, c)
wenzelm@31939
   100
          else if (c + 1 == end || sym < index(c + 1).sym) c
wenzelm@31939
   101
          else bisect(c + 1, b)
wenzelm@31939
   102
        }
wenzelm@31939
   103
        else -1
wenzelm@31939
   104
      }
wenzelm@31939
   105
      val i = bisect(0, end)
wenzelm@31939
   106
      if (i < 0) sym
wenzelm@31939
   107
      else index(i).chr + sym - index(i).sym
wenzelm@31939
   108
    }
wenzelm@31939
   109
  }
wenzelm@31939
   110
wenzelm@31939
   111
wenzelm@34007
   112
  /* recoding text */
wenzelm@27924
   113
wenzelm@31537
   114
  private class Recoder(list: List[(String, String)])
wenzelm@31537
   115
  {
wenzelm@31537
   116
    private val (min, max) =
wenzelm@31537
   117
    {
wenzelm@27937
   118
      var min = '\uffff'
wenzelm@27937
   119
      var max = '\u0000'
wenzelm@27937
   120
      for ((x, _) <- list) {
wenzelm@27937
   121
        val c = x(0)
wenzelm@27937
   122
        if (c < min) min = c
wenzelm@27937
   123
        if (c > max) max = c
wenzelm@27937
   124
      }
wenzelm@27937
   125
      (min, max)
wenzelm@27937
   126
    }
wenzelm@31537
   127
    private val table =
wenzelm@31537
   128
    {
wenzelm@31537
   129
      val table = new jcl.HashMap[String, String]   // reasonably efficient?
wenzelm@27937
   130
      for ((x, y) <- list) table + (x -> y)
wenzelm@27927
   131
      table
wenzelm@27927
   132
    }
wenzelm@31537
   133
    def recode(text: String): String =
wenzelm@31537
   134
    {
wenzelm@27937
   135
      val len = text.length
wenzelm@31537
   136
      val matcher = regex.pattern.matcher(text)
wenzelm@27937
   137
      val result = new StringBuilder(len)
wenzelm@27937
   138
      var i = 0
wenzelm@27937
   139
      while (i < len) {
wenzelm@27937
   140
        val c = text(i)
wenzelm@27937
   141
        if (min <= c && c <= max) {
wenzelm@31939
   142
          matcher.region(i, len).lookingAt
wenzelm@27938
   143
          val x = matcher.group
wenzelm@31537
   144
          result.append(table.get(x) getOrElse x)
wenzelm@27937
   145
          i = matcher.end
wenzelm@27937
   146
        }
wenzelm@27937
   147
        else { result.append(c); i += 1 }
wenzelm@27937
   148
      }
wenzelm@27937
   149
      result.toString
wenzelm@27924
   150
    }
wenzelm@27924
   151
  }
wenzelm@27918
   152
wenzelm@27918
   153
wenzelm@27937
   154
wenzelm@27937
   155
  /** Symbol interpretation **/
wenzelm@27937
   156
wenzelm@34146
   157
  class Interpretation(symbol_decls: List[String])
wenzelm@29569
   158
  {
wenzelm@31537
   159
    /* read symbols */
wenzelm@31537
   160
wenzelm@31537
   161
    private val empty = new Regex("""(?xs) ^\s* (?: \#.* )? $ """)
wenzelm@31537
   162
    private val key = new Regex("""(?xs) (.+): """)
wenzelm@31537
   163
wenzelm@31537
   164
    private def read_decl(decl: String): (String, Map[String, String]) =
wenzelm@31537
   165
    {
wenzelm@31537
   166
      def err() = error("Bad symbol declaration: " + decl)
wenzelm@31537
   167
wenzelm@31537
   168
      def read_props(props: List[String]): Map[String, String] =
wenzelm@31537
   169
      {
wenzelm@31537
   170
        props match {
wenzelm@31537
   171
          case Nil => Map()
wenzelm@31537
   172
          case _ :: Nil => err()
wenzelm@31537
   173
          case key(x) :: y :: rest => read_props(rest) + (x -> y)
wenzelm@31537
   174
          case _ => err()
wenzelm@31537
   175
        }
wenzelm@31537
   176
      }
wenzelm@31537
   177
      decl.split("\\s+").toList match {
wenzelm@34319
   178
        case sym :: props if sym.length > 1 && is_wellformed(sym) => (sym, read_props(props))
wenzelm@34193
   179
        case _ => err()
wenzelm@31537
   180
      }
wenzelm@31537
   181
    }
wenzelm@31537
   182
wenzelm@31537
   183
    private val symbols: List[(String, Map[String, String])] =
wenzelm@34146
   184
      for (decl <- symbol_decls if !empty.pattern.matcher(decl).matches)
wenzelm@31537
   185
        yield read_decl(decl)
wenzelm@31537
   186
wenzelm@31537
   187
wenzelm@31651
   188
    /* misc properties */
wenzelm@31651
   189
wenzelm@34143
   190
    val names: Map[String, String] =
wenzelm@34143
   191
    {
wenzelm@31651
   192
      val name = new Regex("""\\<([A-Za-z][A-Za-z0-9_']*)>""")
wenzelm@31651
   193
      Map((for ((sym @ name(a), _) <- symbols) yield (sym -> a)): _*)
wenzelm@31651
   194
    }
wenzelm@31651
   195
wenzelm@31651
   196
    val abbrevs: Map[String, String] = Map((
wenzelm@31651
   197
      for ((sym, props) <- symbols if props.isDefinedAt("abbrev"))
wenzelm@31651
   198
        yield (sym -> props("abbrev"))): _*)
wenzelm@31651
   199
wenzelm@31651
   200
wenzelm@31537
   201
    /* main recoder methods */
wenzelm@31537
   202
wenzelm@31537
   203
    private val (decoder, encoder) =
wenzelm@31537
   204
    {
wenzelm@31537
   205
      val mapping =
wenzelm@31537
   206
        for {
wenzelm@31537
   207
          (sym, props) <- symbols
wenzelm@31537
   208
          val code =
wenzelm@31537
   209
            try { Integer.decode(props("code")).intValue }
wenzelm@31537
   210
            catch {
wenzelm@31537
   211
              case _: NoSuchElementException => error("Missing code for symbol " + sym)
wenzelm@31537
   212
              case _: NumberFormatException => error("Bad code for symbol " + sym)
wenzelm@31537
   213
            }
wenzelm@31537
   214
          val ch = new String(Character.toChars(code))
wenzelm@34193
   215
        } yield {
wenzelm@34193
   216
          if (code < 128) error("Illegal ASCII code for symbol " + sym)
wenzelm@34193
   217
          else (sym, ch)
wenzelm@34193
   218
        }
wenzelm@31548
   219
      (new Recoder(mapping),
wenzelm@31551
   220
       new Recoder(mapping map { case (x, y) => (y, x) }))
wenzelm@31537
   221
    }
wenzelm@27918
   222
wenzelm@34104
   223
    def decode(text: String): String = decoder.recode(text)
wenzelm@34104
   224
    def encode(text: String): String = encoder.recode(text)
wenzelm@34143
   225
wenzelm@34143
   226
wenzelm@34143
   227
    /* classification */
wenzelm@34143
   228
wenzelm@34147
   229
    private object Decode_Set
wenzelm@34147
   230
    {
wenzelm@34147
   231
      def apply(elems: String*): Set[String] =
wenzelm@34147
   232
      {
wenzelm@34147
   233
        val content = elems.toList
wenzelm@34147
   234
        Set((content ::: content.map(decode)): _*)
wenzelm@34147
   235
      }
wenzelm@34147
   236
    }
wenzelm@34147
   237
wenzelm@34147
   238
    private val letters = Decode_Set(
wenzelm@34143
   239
      "A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M",
wenzelm@34143
   240
      "N", "O", "P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z",
wenzelm@34143
   241
      "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m",
wenzelm@34143
   242
      "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z",
wenzelm@34143
   243
wenzelm@34143
   244
      "\\<A>", "\\<B>", "\\<C>", "\\<D>", "\\<E>", "\\<F>", "\\<G>",
wenzelm@34143
   245
      "\\<H>", "\\<I>", "\\<J>", "\\<K>", "\\<L>", "\\<M>", "\\<N>",
wenzelm@34143
   246
      "\\<O>", "\\<P>", "\\<Q>", "\\<R>", "\\<S>", "\\<T>", "\\<U>",
wenzelm@34143
   247
      "\\<V>", "\\<W>", "\\<X>", "\\<Y>", "\\<Z>", "\\<a>", "\\<b>",
wenzelm@34143
   248
      "\\<c>", "\\<d>", "\\<e>", "\\<f>", "\\<g>", "\\<h>", "\\<i>",
wenzelm@34143
   249
      "\\<j>", "\\<k>", "\\<l>", "\\<m>", "\\<n>", "\\<o>", "\\<p>",
wenzelm@34143
   250
      "\\<q>", "\\<r>", "\\<s>", "\\<t>", "\\<u>", "\\<v>", "\\<w>",
wenzelm@34143
   251
      "\\<x>", "\\<y>", "\\<z>",
wenzelm@34143
   252
wenzelm@34143
   253
      "\\<AA>", "\\<BB>", "\\<CC>", "\\<DD>", "\\<EE>", "\\<FF>",
wenzelm@34143
   254
      "\\<GG>", "\\<HH>", "\\<II>", "\\<JJ>", "\\<KK>", "\\<LL>",
wenzelm@34143
   255
      "\\<MM>", "\\<NN>", "\\<OO>", "\\<PP>", "\\<QQ>", "\\<RR>",
wenzelm@34143
   256
      "\\<SS>", "\\<TT>", "\\<UU>", "\\<VV>", "\\<WW>", "\\<XX>",
wenzelm@34143
   257
      "\\<YY>", "\\<ZZ>", "\\<aa>", "\\<bb>", "\\<cc>", "\\<dd>",
wenzelm@34143
   258
      "\\<ee>", "\\<ff>", "\\<gg>", "\\<hh>", "\\<ii>", "\\<jj>",
wenzelm@34143
   259
      "\\<kk>", "\\<ll>", "\\<mm>", "\\<nn>", "\\<oo>", "\\<pp>",
wenzelm@34143
   260
      "\\<qq>", "\\<rr>", "\\<ss>", "\\<tt>", "\\<uu>", "\\<vv>",
wenzelm@34143
   261
      "\\<ww>", "\\<xx>", "\\<yy>", "\\<zz>",
wenzelm@34143
   262
wenzelm@34143
   263
      "\\<alpha>", "\\<beta>", "\\<gamma>", "\\<delta>", "\\<epsilon>",
wenzelm@34143
   264
      "\\<zeta>", "\\<eta>", "\\<theta>", "\\<iota>", "\\<kappa>",
wenzelm@34143
   265
      "\\<mu>", "\\<nu>", "\\<xi>", "\\<pi>", "\\<rho>", "\\<sigma>",
wenzelm@34143
   266
      "\\<tau>", "\\<upsilon>", "\\<phi>", "\\<chi>", "\\<psi>",
wenzelm@34143
   267
      "\\<omega>", "\\<Gamma>", "\\<Delta>", "\\<Theta>", "\\<Lambda>",
wenzelm@34143
   268
      "\\<Xi>", "\\<Pi>", "\\<Sigma>", "\\<Upsilon>", "\\<Phi>",
wenzelm@34143
   269
      "\\<Psi>", "\\<Omega>",
wenzelm@34143
   270
wenzelm@34143
   271
      "\\<^isub>", "\\<^isup>")
wenzelm@34143
   272
wenzelm@34147
   273
    private val blanks =
wenzelm@34147
   274
      Decode_Set(" ", "\t", "\n", "\u000B", "\f", "\r", "\\<spacespace>", "\\<^newline>")
wenzelm@34147
   275
wenzelm@34147
   276
    private val sym_chars =
wenzelm@34147
   277
      Set("!", "#", "$", "%", "&", "*", "+", "-", "/", "<", "=", ">", "?", "@", "^", "_", "|", "~")
wenzelm@34143
   278
wenzelm@34143
   279
    def is_letter(sym: String): Boolean = letters.contains(sym)
wenzelm@34147
   280
    def is_digit(sym: String): Boolean = sym.length == 1 && '0' <= sym(0) && sym(0) <= '9'
wenzelm@34143
   281
    def is_quasi(sym: String): Boolean = sym == "_" || sym == "'"
wenzelm@34147
   282
    def is_letdig(sym: String): Boolean = is_letter(sym) || is_digit(sym) || is_quasi(sym)
wenzelm@34143
   283
    def is_blank(sym: String): Boolean = blanks.contains(sym)
wenzelm@34147
   284
    def is_symbolic_char(sym: String): Boolean = sym_chars.contains(sym)
wenzelm@34147
   285
    def is_symbolic(sym: String): Boolean = sym.startsWith("\\<") && !sym.startsWith("\\<^")
wenzelm@27918
   286
  }
wenzelm@27901
   287
}